Re: [PATCH][PPC32] Xilinx ML300 board support (very basic)

[Andrei Konovalov <akonovalov@ru.mvista.com>]

Jon Masters wrote:
> On Fri, 08 Oct 2004 21:23:01 +0400, Andrei Konovalov
> <akonovalov@ru.mvista.com> wrote:
> 
> 
>>Adds minimal Xilinx ML300 board support (enough to boot with ramdisk).
>>The only peripheral devices supported are 16x50 compatible UARTs.
>>
>>Signed-off-by: Andrei Konovalov <akonovalov@ru.mvista.com>
> 
> 
> I'll test this if I get chance soon.
> 

Great!

> 
>>        /*
>>-        * At one point, we were getting machine checks.  Linux was not
>>-        * invalidating the data cache before it was enabled.  The
>>-        * following code was added to do that.  Soon after we had done
>>-        * that, we found the real reasons for the machine checks.  I've
>>-        * run the kernel a few times with the following code
>>-        * temporarily removed without any apparent problems.  However,
>>-        * I objdump'ed the kernel and boot code and found out that
>>-        * there were no other dccci's anywhere, so I put the code back
>>-        * in and have been reluctant to remove it.  It seems safer to
>>-        * just leave it here.
>>+        * Invalidate the data cache if the data cache is turned off.
>>+        * - The 405 core does not invalidate the data cache on power-up
>>+        *   or reset but does turn off the data cache. We cannot assume
>>+        *   that the cache contents are valid.
>>+        * - If the data cache is turned on this must have been done by
>>+        *   a bootloader and we assume that the cache contents are
>>+        *   valid.
>>         */
> 
> 
> Er...the caveat about the MSR is still worth noting - I had this
> problem and for a while couldn't figure out why the fsck the board
> wouldn't come up with cacheing enabled (I've written patches for
> debugging that turn off various cacheing by modifying the pte macros,
> etc.) and *still* believe we're not clean from unwanted machine checks
> here. Just FYI.
> 

The code which follows the comment above works for me with both ML300
and the Memec 2VP7 boards I have.
Do you mean that some boards (or FPGA designs) need cacheing to be
disabled completely?

If this is the case, and you have the patch that handles this, I would
welcome that patch to follow mine.

Is there any chance for me to reproduce that problem on my site with
the hardware I have (1. ML300; 2. Memec 2VP7 board plus the COMM and the
System ACE modules)?

> 
...
> 
>>+*     This program is free software; you can redistribute it and/or modify it
>>+*     under the terms of the GNU General Public License as published by the
>>+*     Free Software Foundation; either version 2 of the License, or (at your
>>+*     option) any later version.
> 
> 
> ...and then:
> 
> 
>>+*     Xilinx hardware products are not intended for use in life support
>>+*     appliances, devices, or systems. Use in such applications is
>>+*     expressly prohibited.
> 
> 
> I've always loved it how Xilinx GPL disclaimers cunningly aren't. They
> might want to wake up and smell the coffee at some point.

I know this doesn't look good.
If it *does* prevent this code from getting into the kernel tree I could
try contacting Xilinx (doubt it will help though - they have already
changed the copyrights once to be more community friendly).

> 
> Andrei, you'll want to modify the ML300 xparameters stuff to allow
> it's location to be specified by a parameter. People who want to use
> (ewww, spit) autogenerated Xilinx xparameters.h from their EDK will
> probably also want to choose where it lives.
> 

Why renaming someone's autogenerated xparameters.h into xparameters_my_ml300.h
and placing it into arch/ppc/platforms/4xx/xparameters/,
defining CONFIG_XILINX_MY_ML300, and adding
few lines to include/asm-ppc/xparameters.h like:

#if defined(CONFIG_XILINX_ML300)
#include <platforms/4xx/xparameters/xparameters_ml300.h>
#endif

+#if defined(CONFIG_XILINX_MY_ML300)
+#include <platforms/4xx/xparameters/xparameters_my_ml300.h>
+#endif

doesn't work in your oppinion?

> 
...
> 
> Jon.
> 
> 

Thanks,
Andrei